Milad Azarbad

Bio: Milad Azarbad is an academic researcher from University of Mazandaran. The author has contributed to research in topics: Wavelet & Image segmentation. The author has an hindex of 5, co-authored 9 publications receiving 68 citations.

New Neural Network-based Approaches for GPS GDOP Classification based on Neuro-Fuzzy Inference System, Radial Basis Function, and Improved Bee Algorithm

Abstract: GDOP stands as a relevant measure of positioning accuracy.We propose hybrid intelligent methods, namely ANFIS, improved ANFIS, and RBF, for GPS GDOP classification.Bee algorithm (BA) and improved BA are proposed for finding the optimum radius vector of the ANFIS.To enhance the classification accuracy, PCA is utilized. Global positioning system (GPS) is the most widely used military and commercial positioning tool for real-time navigation and location. Geometric dilution of precision (GDOP) stands as a relevant measure of positioning accuracy and consequently, the performance quality of the GPS positioning algorithm. Since the calculation of GPS GDOP has a time and power burden that involves complicated transformation and inversion of measurement matrices, in this paper we propose hybrid intelligent methods, namely adaptive neuro-fuzzy inference system (ANFIS), improved ANFIS, and radial basis function (RBF), for GPS GDOP classification. Through investigation it is verified that the ANFIS is a high performance and valuable classifier. In the ANFIS training, the radius vector has very important role for its recognition accuracy. Therefore, in the optimization module, bee algorithm (BA) is proposed for finding the optimum vector of radius. In order to improve the performance of the proposed method, a new improvement for the BA is used. In addition, to enhance the accuracy of the method, principal component analysis (PCA) is utilized as a pre-processing step. Experimental results clearly indicate that the proposed intelligent methods have high classification accuracy rates comparing with conventional ones.

Automatic Recognition of Digital Communication Signal

Abstract: Automatic recognition of digital communication signals has seen increasing demands nowadays in various applications. This paper investigates the design of high efficient system for classification of the digital communication signals. The system includes two main modules: feature extraction and classification. In the feature extraction module we have used a novel balanced combination of the higher order moments (up to eighth), higher order cumulants (up to eighth) and spectral characteristics. In the classifier we have investigated the performances of the radial basis function (RBF) neural network, probability neural network (PNN) and multilayer perceptron (MLP) neural network. Then we have compared these systems. Experimental results show the proposed systems have high percentage of correct classification to discriminate the different types of digital signals even at very low SNRs.

A time-frequency approach for EEG signal segmentation

Abstract: The record of human brain neural activities, namely electroencephalogram (EEG), is known to be nonstationary in general. In addition, the human head is a non-linear medium for such signals. In many applications, it is useful to divide the EEGs into segments in which the signals can be considered stationary. Here, Hilbert-Huang Transform (HHT), as an effective tool in signal processing is applied since unlike the traditional time-frequency approaches, it exploits the non-linearity of the medium and nonstationarity of the EEG signals. In addition, we use Singular Spectrum Analysis (SSA) in the pre-processing step as an effective noise removal approach. By using synthetic and real EEG signals, the proposed method is compared with Wavelet Generalized Likelihood Ratio (WGLR) algorithm as a well-known signal segmentation method. The simulation results indicate the performance superiority of the proposed method.

Brain tissue segmentation using an unsupervised clustering technique based on PSO algorithm

Abstract: Image thresholding is an important technique for image processing and pattern recognition. Several thresholding techniques have been proposed in the literature. In this paper for segmentation of magnetic resonance images, a novel method using a combination of the multilevel thresholding algorithm and the hierarchical evolutionary algorithm (HEA) is proposed. The HEA can be viewed as a variant of conventional genetic algorithms. The proposed technique is based on the participle swarm optimization (PSO) and, in fact, is an unsupervised clustering method based on an automatic multilevel thresholding approach. One advantage of the proposed method is that the number of clusters in the given image does not need to be known in advance. We evaluate and validate performance of the proposed method using simulation studies. The simulation results show that the accuracy of the proposed method is about 96%.

ECG compression using wavelet transform and three-level quantization

Abstract: Electrocardiogram signal is a very useful source of information for physicians in diagnosing heart abnormalities. With the increasing use of ECG in heart diagnosis, such as 24 hours monitoring or in ambulatory monitoring systems, the volume of ECG data that should be stored or transmitted, has greatly increased. This paper presents an efficient technique for compression of electrocardiogram (ECG) signals. In this technique, it is used the three level of quantization for thresholding. Then for encoding of samples we use an embedded of zero-tree wavelet (EZW) method and Huffman algorithms. The EZW algorithm allows an optimal data compression for a target bit rate and appeared superior to other wavelet-based ECG coders. Also, we have implemented different types of wavelet for compression of the ECG records from MIT-BIH database and compared their performances in the proposed algorithm. We have evaluated the effect of signal length and iterations of quantization on the compression ratio (CR). The correlation coefficient and percentage root mean square difference (PRD) is also measured. Simulation results show that the proposed method has good performance for especial types of wavelet transform. In general existing measures show how much reconstructed signal is similar to the original one.

An adaptive and hybrid artificial bee colony algorithm (aABC) for ANFIS training

Abstract: Display Omitted We propose an Adaptive and Hybrid Artificial Bee Colony (aABC) algorithm based on ABC algorithm to train ANFIS.We use crossover rate and adaptivity coefficient in aABC algorithm, Unlike the standard ABC algorithm.We gain the rapid convergence feature with the usage of arithmetic crossover.We apply the proposed model on two different problem groups: solution of numerical optimization problems and training ANFIS.The obtained results are compared with the other algorithms which are commonly used in the literature. In this study, we propose an Adaptive and Hybrid Artificial Bee Colony (aABC) algorithm to train ANFIS. Unlike the standard ABC algorithm, two new parameters are utilized in the solution search equation. These are arithmetic crossover rate and adaptivity coefficient. aABC algorithm gains the rapid convergence feature with the usage of arithmetic crossover and it is applied on two different problem groups and its performance is measured. Firstly, it is performed over 10 numerical benchmark functions. The results show that aABC algorithm is more efficient than standard ABC algorithm. Secondly, ANFIS is trained by using aABC algorithm to identify the nonlinear dynamic systems. Each application begins with the randomly selected initial population and then average RMSE is obtained. For four examples considered in ANFIS training, train error values are respectively computed as 0.0344, 0.0232, 0.0152 and 0.0205. Also, test error values for these examples are respectively found as 0.0255, 0.0202, 0.0146 and 0.0295. Although it varies according to the examples, performance increase between 4.51% and 33.33% occurs. Additionally, it is seen that aABC algorithm converges bettter than ABC algorithm in the all examples. The obtained results are compared with the neuro-fuzzy based approaches which are commonly used in the literature and it is seen that the proposed ABC variant can be efficiently used for ANFIS training.

Evaluation of current dental radiographs segmentation approaches in computer-aided applications

Abstract: With a wide variety researches on Image segmentation techniques in biomedical and bioinformatics area, it is important to analyze the performance of these approaches in specific problems. Image segmentation is one of the most significant processes of dental X-ray image analysis. Therefore, to obtain the proper result, it is required to perform the accurate and efficient segmentation approach which proved itself in the aspect of X-ray image segmentation. The aim of this review paper is to understand the different image segmentation approaches which have been used for dental X-ray image analysis over the past studies. In this paper, different available approaches of dental X-ray image segmentation, reviewed and their advantages, disadvantages, and limitations are discussed.